Extended Data Fig. 5: MitoPerturb-seq identifies mtDNA depletion following targeted gene perturbation.
(A) Cells assigned to individual gRNA groups (NT, Tfam, Opa1 and Polg UMAPs shown in Fig. 2f-i). (B) Relative enrichment of individual gRNA groups in cluster 4 (low mtDNA CN) compared to all other clusters. Target genes highlighted red are over-represented and blue under-represented in cluster 4. Two-sided chi-squared tests with multiple testing correction (Benjamini-Hochberg) and odds ratio calculations. All adjusted p-values shown; *p < 0.05, **p < 0.01, ***p < 0.001. NB: Scaled colouring is capped at p = 1×10−20 for clarity, exact p-values are on plots. (C-F) Per-cell nuclear ATAC fragment count (C) mtDNA coverage (D), mtDNA transcript levels (E) and heteroplasmy (F) for each gRNA group, including controls. Each violin represents one of three gRNAs targeting the corresponding gene. Pairwise two-sided Welch’s t-tests with multiple testing correction (Bonferroni), p-values in figure, all values < 0.05 shown. Black horizontal lines indicate median (C-E) or mean (F). Plots include all cells in the final integrated dataset, without additional filtering. (G,H) Per-cell mtDNA coverage of the WT (G) and mutant (H) allele (no depth cut-off applied). (I) Group size required to confidently identify mean heteroplasmy shifts at power >0.8 and p < 0.05. A mean group size of 179 would suffice to identify mean heteroplasmy shifts of 3.5-4%, assuming all cells shifted in the same direction. (J) Observed heteroplasmy distributions for Tfam, Opa1 and Polg KO compared to simulated distributions generated by binomial downsampling of control cells to matched read depths (n = 5,000). Boxes 25th-75th percentiles, mid-line median, whiskers extend to most extreme datapoints within 1.5 x IQR from hinges, all datapoints plotted. (K) Relationship between mtDNA depth and heteroplasmy variance. Single cells aggregated into read-depth bins of equal sample size to calculate local variance. Gray points without mtDNA depth reduction (all gRNAs excluding Tfam, Polg, and Opa1); coloured points represent Tfam, Polg, and Opa1 perturbation groups. The gray dashed line represents the 95th percentile fit from quantile regression on control cells (95% of control cells have technical variance at or below this line at a given depth). The gray ribbon represents bootstrapped 95% confidence intervals for the expected variance range modelled on the non-depleted population.
